Building at the intersection of complex engineering systems, local AI pipelines, and financial technology.
Multi-disciplinary execution — bridging deep cyber-physical system architecture, model-based automation workflows, and high-ticket fintech monetization logic. I write what I build, and I build in public.
Three disciplines, one operating system.
Cyber-Physical Systems
Embedded firmware, model-based design, and sensor fusion across constrained edge environments.
Local AI Pipelines
Self-hosted inference, RAG over private corpora, and orchestration on metal you actually own.
Fintech Architecture
Idempotent ledgers, programmable payouts, and high-ticket payment frameworks engineered for trust.
The metal beneath the writing.
What's on the bench, right now.
Local-First RAG Pipeline on ARM Edge Cluster
Wiring Ollama + Qdrant on a 4-node ARM edge mesh, fronted by a Portainer-managed gateway. Target: sub-200ms semantic retrieval over a 12GB private corpus.
- 01Model-based testbench for cyber-physical sensor fusion (MATLAB → Docker)
- 02Fintech: programmable payouts API with idempotent ledger primitives
- 03Self-hosted observability stack: Grafana + Loki + Tempo on Mac mini
- 04ESP32-S3 edge inference: distilled vision model under 8MB
Most recent dispatches.
Why I Run My AI Stack Locally (And You Probably Should Too)
A pragmatic breakdown of the latency, privacy, and unit-economics arguments for running inference on metal you actually own.
A Portainer-Driven Home Lab Without the Yak Shave
The single-file compose stack I use to bring a fresh ARM node from BIOS to managed-in-Portainer in under 20 minutes.
When the MBA Meets the Embedded Systems Engineer
The most useful unlock of business school wasn't strategy frameworks — it was learning to price the cost of being wrong.